1IRCCS SDN, Naples;2Department of Bioengineering, Politecnico di Milano, Milan;3Institute of Biostructure and Bioimaging,National Research Council, Naples;4Department of Advanced Biomedical Sciences, University Federico Ii, Naples, Italy
References: Not available.
ENHANCED VESSELNESS MAPPING TO DETECT SMALL CEREBRAL VEINS ON SWI IMAGES
Background
: Cerebral veins are visualized by Susceptibility Weighted Imaging (SWI), which may highlight vascular abnormalities in different cerebral diseases. Assessing the vascular anatomy in a 3D computer model by a vascular extraction algorithm may improve the study of structural and functional abnormalities in vascular diseases.
Objectives
: To implement a modified version of Vessel Enhancing Diffusion (VED) filter and test its application to SWI datasets at different flip angles.
Methods
: A multi-scale (0.1 to 1 mm) vessel enhancement filter was applied to several 3D SWI datasets acquired by a 3T scanner (FAs=[3~20]°, TR=31 ms, TE=22.14 ms, voxel size=0.5x0.5x1 mm3).
For each scale and voxel the Hessian tensor was calculated and a vesselness estimate was computed on the basis of the Hessian eigenvalues,to distinguish tubular from non-tubular structures. The voxel-byvoxel maximum of these estimates was evolved according to a scheme of anisotropic diffusion highly restricted when perpendicular to tubular
structures (VED). To improve the VED filtering and obtain high-SNR continuous 3D model of the veins, the diffusion time was set to 0.2, the sensitivity s to 3 and, at each diffusion step, the vesselness
function was thresholded, suppressing all values below 0.02.
Results:
The visual inspection of the vesselness maps obtained varying the SWI flip angle suggests that the better compromise between sensitivity to vessels and specificity against CSF-filled sulci was
reached using flip angles close to the Ernst angle of gray matter (~12°for the given TR).
Conclusions
: Dynamical thresholding of the vesselness function within the diffusion evolution provides an increased specificity in vessel visualization that, combined with a proper choice of the acquisition flip angle, allows the detection of deep medullary veins barely visible on the original SWI.
ENHANCED VESSELNESS MAPPING TO DETECT SMALL CEREBRAL VEINS ON SWI IMAGES